X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=lexer.c;h=1f0c23d46b004cb0cea3ed84a7689f3c9c55c5df;hb=34a560b81b8899995ff67421650edcad53efb4d4;hp=9ff748a32867c3fa67c8a18de7bd7a42f2531c40;hpb=35b6bfa9d3d3cbaa12c753901d53751a3ea7c0af;p=cparser diff --git a/lexer.c b/lexer.c index 9ff748a..1f0c23d 100644 --- a/lexer.c +++ b/lexer.c @@ -1,101 +1,134 @@ +/* + * This file is part of cparser. + * Copyright (C) 2007-2009 Matthias Braun + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA + * 02111-1307, USA. + */ #include +#include "adt/strutil.h" +#include "input.h" +#include "diagnostic.h" #include "lexer.h" +#include "symbol_t.h" #include "token_t.h" #include "symbol_table_t.h" #include "adt/error.h" #include "adt/strset.h" #include "adt/util.h" +#include "types.h" +#include "type_t.h" +#include "parser.h" +#include "warning.h" +#include "lang_features.h" #include #include #include +#include #include -//#define DEBUG_CHARS -#define MAX_PUTBACK 3 - -static int c; -token_t lexer_token; -static FILE *input; -static char buf[1024 + MAX_PUTBACK]; -static const char *bufend; -static const char *bufpos; -static strset_t stringset; - -static void error_prefix_at(const char *input_name, unsigned linenr) -{ - fprintf(stderr, "%s:%u: Error: ", input_name, linenr); -} +#ifndef _WIN32 +#include +#endif -static void error_prefix(void) +#define MAX_PUTBACK 16 // 3 would be enough, but 16 gives a nicer alignment +#define BUF_SIZE 1024 + +static input_t *input; +static utf32 input_buf[BUF_SIZE + MAX_PUTBACK]; +static const utf32 *bufpos; +static const utf32 *bufend; +static utf32 c; +static source_position_t lexer_pos; +token_t lexer_token; +static symbol_t *symbol_L; +static strset_t stringset; +bool allow_dollar_in_symbol = true; + +/** + * Prints a parse error message at the current token. + * + * @param msg the error message + */ +static void parse_error(const char *msg) { - error_prefix_at(lexer_token.source_position.input_name, - lexer_token.source_position.linenr); + errorf(&lexer_pos, "%s", msg); } -static void parse_error(const char *msg) +/** + * Prints an internal error message at the current token. + * + * @param msg the error message + */ +static NORETURN internal_error(const char *msg) { - error_prefix(); - fprintf(stderr, "%s\n", msg); + internal_errorf(&lexer_pos, "%s", msg); } static inline void next_real_char(void) { - bufpos++; - if(bufpos >= bufend) { - size_t s = fread(buf + MAX_PUTBACK, 1, sizeof(buf) - MAX_PUTBACK, - input); - if(s == 0) { + assert(bufpos <= bufend); + if (bufpos >= bufend) { + size_t n = decode(input, input_buf+MAX_PUTBACK, BUF_SIZE); + if (n == 0) { c = EOF; return; } - bufpos = buf + MAX_PUTBACK; - bufend = buf + MAX_PUTBACK + s; + bufpos = input_buf + MAX_PUTBACK; + bufend = bufpos + n; } - c = *(bufpos); + c = *bufpos++; + ++lexer_pos.colno; } -static inline void put_back(int pc) +/** + * Put a character back into the buffer. + * + * @param pc the character to put back + */ +static inline void put_back(utf32 const pc) { - assert(bufpos >= buf); - assert(bufpos < buf+MAX_PUTBACK || *bufpos == pc); - - char *p = buf + (bufpos - buf); - *p = pc; - - /* going backwards in the buffer is legal as long as it's not more often - * than MAX_PUTBACK */ - bufpos--; - -#ifdef DEBUG_CHARS - printf("putback '%c'\n", pc); -#endif + *(--bufpos - input_buf + input_buf) = pc; + --lexer_pos.colno; } static inline void next_char(void); -#define MATCH_NEWLINE(code) \ - case '\r': \ - next_char(); \ - if(c == '\n') { \ - next_char(); \ - } \ - lexer_token.source_position.linenr++; \ - code; \ - case '\n': \ - next_char(); \ - lexer_token.source_position.linenr++; \ - code; - -#define eat(c_type) do { assert(c == c_type); next_char(); } while(0) +#define NEWLINE \ + '\r': \ + next_char(); \ + if (c == '\n') { \ + case '\n': \ + next_char(); \ + } \ + lexer_pos.lineno++; \ + lexer_pos.colno = 1; \ + goto newline; \ + newline // Let it look like an ordinary case label. + +#define eat(c_type) (assert(c == c_type), next_char()) static void maybe_concat_lines(void) { eat('\\'); - switch(c) { - MATCH_NEWLINE(return;) + switch (c) { + case NEWLINE: + return; default: break; @@ -105,29 +138,32 @@ static void maybe_concat_lines(void) c = '\\'; } +/** + * Set c to the next input character, ie. + * after expanding trigraphs. + */ static inline void next_char(void) { next_real_char(); -#if 0 /* filter trigraphs */ - if(UNLIKELY(c == '\\')) { + if (UNLIKELY(c == '\\')) { maybe_concat_lines(); - goto end_of_next_char; + return; } - if(LIKELY(c != '?')) - goto end_of_next_char; + if (LIKELY(c != '?')) + return; next_real_char(); - if(LIKELY(c != '?')) { + if (LIKELY(c != '?')) { put_back(c); c = '?'; - goto end_of_next_char; + return; } next_real_char(); - switch(c) { + switch (c) { case '=': c = '#'; break; case '(': c = '['; break; case '/': c = '\\'; maybe_concat_lines(); break; @@ -138,28 +174,19 @@ static inline void next_char(void) case '>': c = '}'; break; case '-': c = '~'; break; default: - put_back('?'); put_back(c); + put_back('?'); c = '?'; break; } - -end_of_next_char: -#endif - (void) maybe_concat_lines; -#ifdef DEBUG_CHARS - printf("nchar '%c'\n", c); -#else - ; -#endif } -#define SYMBOL_CHARS \ +#define SYMBOL_CHARS_WITHOUT_E_P \ + '$': if (!allow_dollar_in_symbol) goto dollar_sign; \ case 'a': \ case 'b': \ case 'c': \ case 'd': \ - case 'e': \ case 'f': \ case 'g': \ case 'h': \ @@ -170,7 +197,6 @@ end_of_next_char: case 'm': \ case 'n': \ case 'o': \ - case 'p': \ case 'q': \ case 'r': \ case 's': \ @@ -185,7 +211,6 @@ end_of_next_char: case 'B': \ case 'C': \ case 'D': \ - case 'E': \ case 'F': \ case 'G': \ case 'H': \ @@ -196,7 +221,6 @@ end_of_next_char: case 'M': \ case 'N': \ case 'O': \ - case 'P': \ case 'Q': \ case 'R': \ case 'S': \ @@ -207,10 +231,20 @@ end_of_next_char: case 'X': \ case 'Y': \ case 'Z': \ - case '_': + case '_' + +#define SYMBOL_CHARS_E_P \ + 'E': \ + case 'P': \ + case 'e': \ + case 'p' + +#define SYMBOL_CHARS \ + SYMBOL_CHARS_WITHOUT_E_P: \ + case SYMBOL_CHARS_E_P #define DIGITS \ - case '0': \ + '0': \ case '1': \ case '2': \ case '3': \ @@ -219,304 +253,292 @@ end_of_next_char: case '6': \ case '7': \ case '8': \ - case '9': + case '9' -static void parse_symbol(void) +static bool is_universal_char_valid(utf32 const v) { - symbol_t *symbol; - char *string; + /* C11 §6.4.3:2 */ + if (v < 0xA0U && v != 0x24 && v != 0x40 && v != 0x60) + return false; + if (0xD800 <= v && v <= 0xDFFF) + return false; + return true; +} - obstack_1grow(&symbol_obstack, c); - next_char(); +static int digit_value(utf32 digit); - while(1) { - switch(c) { - DIGITS - SYMBOL_CHARS - obstack_1grow(&symbol_obstack, c); +static utf32 parse_universal_char(unsigned const n_digits) +{ + utf32 v = 0; + for (unsigned k = n_digits; k != 0; --k) { + if (isxdigit(c)) { + v = 16 * v + digit_value(c); next_char(); + } else { + errorf(&lexer_pos, "short universal character name, expected %u more digits", k); break; - - default: - goto end_symbol; } } - -end_symbol: - obstack_1grow(&symbol_obstack, '\0'); - - string = obstack_finish(&symbol_obstack); - symbol = symbol_table_insert(string); - - lexer_token.type = symbol->ID; - lexer_token.v.symbol = symbol; - - if(symbol->string != string) { - obstack_free(&symbol_obstack, string); + if (!is_universal_char_valid(v)) { + errorf(&lexer_pos, "\\%c%0*X is not a valid universal character name", n_digits == 4 ? 'u' : 'U', (int)n_digits, v); } + return v; } -static void parse_integer_suffix(void) +static bool is_universal_char_valid_identifier(utf32 const v) { - if(c == 'U' || c == 'U') { - /* TODO do something with the suffixes... */ - next_char(); - if(c == 'L' || c == 'l') { - next_char(); - if(c == 'L' || c == 'l') { - next_char(); - } - } - } else if(c == 'l' || c == 'L') { - next_char(); - if(c == 'l' || c == 'L') { - next_char(); - if(c == 'u' || c == 'U') { - next_char(); - } - } else if(c == 'u' || c == 'U') { - next_char(); - } - } + /* C11 Annex D.1 */ + if ( v == 0x000A8) return true; + if ( v == 0x000AA) return true; + if ( v == 0x000AD) return true; + if ( v == 0x000AF) return true; + if (0x000B2 <= v && v <= 0x000B5) return true; + if (0x000B7 <= v && v <= 0x000BA) return true; + if (0x000BC <= v && v <= 0x000BE) return true; + if (0x000C0 <= v && v <= 0x000D6) return true; + if (0x000D8 <= v && v <= 0x000F6) return true; + if (0x000F8 <= v && v <= 0x000FF) return true; + if (0x00100 <= v && v <= 0x0167F) return true; + if (0x01681 <= v && v <= 0x0180D) return true; + if (0x0180F <= v && v <= 0x01FFF) return true; + if (0x0200B <= v && v <= 0x0200D) return true; + if (0x0202A <= v && v <= 0x0202E) return true; + if (0x0203F <= v && v <= 0x02040) return true; + if ( v == 0x02054) return true; + if (0x02060 <= v && v <= 0x0206F) return true; + if (0x02070 <= v && v <= 0x0218F) return true; + if (0x02460 <= v && v <= 0x024FF) return true; + if (0x02776 <= v && v <= 0x02793) return true; + if (0x02C00 <= v && v <= 0x02DFF) return true; + if (0x02E80 <= v && v <= 0x02FFF) return true; + if (0x03004 <= v && v <= 0x03007) return true; + if (0x03021 <= v && v <= 0x0302F) return true; + if (0x03031 <= v && v <= 0x0303F) return true; + if (0x03040 <= v && v <= 0x0D7FF) return true; + if (0x0F900 <= v && v <= 0x0FD3D) return true; + if (0x0FD40 <= v && v <= 0x0FDCF) return true; + if (0x0FDF0 <= v && v <= 0x0FE44) return true; + if (0x0FE47 <= v && v <= 0x0FFFD) return true; + if (0x10000 <= v && v <= 0x1FFFD) return true; + if (0x20000 <= v && v <= 0x2FFFD) return true; + if (0x30000 <= v && v <= 0x3FFFD) return true; + if (0x40000 <= v && v <= 0x4FFFD) return true; + if (0x50000 <= v && v <= 0x5FFFD) return true; + if (0x60000 <= v && v <= 0x6FFFD) return true; + if (0x70000 <= v && v <= 0x7FFFD) return true; + if (0x80000 <= v && v <= 0x8FFFD) return true; + if (0x90000 <= v && v <= 0x9FFFD) return true; + if (0xA0000 <= v && v <= 0xAFFFD) return true; + if (0xB0000 <= v && v <= 0xBFFFD) return true; + if (0xC0000 <= v && v <= 0xCFFFD) return true; + if (0xD0000 <= v && v <= 0xDFFFD) return true; + if (0xE0000 <= v && v <= 0xEFFFD) return true; + return false; } -static void parse_floating_suffix(void) +static bool is_universal_char_valid_identifier_start(utf32 const v) { - switch(c) { - /* TODO: do something usefull with the suffixes... */ - case 'f': - case 'F': - case 'l': - case 'L': - next_char(); - break; - default: - break; - } + /* C11 Annex D.2 */ + if (0x0300 <= v && v <= 0x036F) return false; + if (0x1DC0 <= v && v <= 0x1DFF) return false; + if (0x20D0 <= v && v <= 0x20FF) return false; + if (0xFE20 <= v && v <= 0xFE2F) return false; + return true; } -static void parse_number_hex(void) +/** + * Read a symbol from the input and build + * the lexer_token. + */ +static void parse_symbol(void) { - assert(c == 'x' || c == 'X'); - next_char(); + while (true) { + switch (c) { + case DIGITS: + case SYMBOL_CHARS: + obstack_1grow(&symbol_obstack, (char) c); + next_char(); + break; - if (!isdigit(c) && - !('A' <= c && c <= 'F') && - !('a' <= c && c <= 'f')) { - parse_error("premature end of hex number literal"); - lexer_token.type = T_ERROR; - return; - } + case '\\': + next_char(); + switch (c) { + { + unsigned n; + case 'U': n = 8; goto universal; + case 'u': n = 4; goto universal; +universal: + next_char(); + utf32 const v = parse_universal_char(n); + if (!is_universal_char_valid_identifier(v)) { + if (is_universal_char_valid(v)) { + errorf(&lexer_pos, "universal character \\%c%0*X is not valid in an identifier", n == 4 ? 'u' : 'U', (int)n, v); + } + } else if (obstack_object_size(&symbol_obstack) == 0 && !is_universal_char_valid_identifier_start(v)) { + errorf(&lexer_pos, "universal character \\%c%0*X is not valid as start of an identifier", n == 4 ? 'u' : 'U', (int)n, v); + } else { + obstack_grow_utf8(&symbol_obstack, v); + } + break; + } - int value = 0; - while(1) { - if (isdigit(c)) { - value = 16 * value + c - '0'; - } else if ('A' <= c && c <= 'F') { - value = 16 * value + c - 'A' + 10; - } else if ('a' <= c && c <= 'f') { - value = 16 * value + c - 'a' + 10; - } else { - parse_integer_suffix(); + default: + put_back(c); + c = '\\'; + goto end_symbol; + } - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = value; - return; + default: +dollar_sign: + goto end_symbol; } - next_char(); } - if(c == '.' || c == 'p' || c == 'P') { - next_char(); - panic("Hex floating point numbers not implemented yet"); - } -} +end_symbol: + obstack_1grow(&symbol_obstack, '\0'); -static void parse_number_oct(void) -{ - int value = 0; - while(c >= '0' && c <= '7') { - value = 8 * value + c - '0'; - next_char(); - } - if (c == '8' || c == '9') { - parse_error("invalid octal number"); - lexer_token.type = T_ERROR; - return; - } + char *string = obstack_finish(&symbol_obstack); + symbol_t *symbol = symbol_table_insert(string); - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = value; + lexer_token.kind = symbol->ID; + lexer_token.base.symbol = symbol; - parse_integer_suffix(); + if (symbol->string != string) { + obstack_free(&symbol_obstack, string); + } } -static void parse_floatingpoint_exponent(long double value) +static string_t sym_make_string(string_encoding_t const enc) { - unsigned int expo = 0; - long double factor = 10.; - - if(c == '-') { - next_char(); - factor = 0.1; - } else if(c == '+') { - next_char(); - } - - while(c >= '0' && c <= '9') { - expo = 10 * expo + (c - '0'); - next_char(); - } + obstack_1grow(&symbol_obstack, '\0'); + size_t const len = obstack_object_size(&symbol_obstack) - 1; + char *const string = obstack_finish(&symbol_obstack); - while(1) { - if(expo & 1) - value *= factor; - expo >>= 1; - if(expo == 0) - break; - factor *= factor; + /* TODO hash */ +#if 0 + const char *result = strset_insert(&stringset, concat); + if (result != concat) { + obstack_free(&symbol_obstack, concat); } - - lexer_token.type = T_FLOATINGPOINT; - lexer_token.v.floatvalue = value; - - parse_floating_suffix(); +#else + const char *result = string; +#endif + return (string_t){ result, len, enc }; } -static void parse_floatingpoint_fract(int integer_part) +static void parse_pp_number(void) { - long double value = integer_part; - long double factor = 1.; - - while(c >= '0' && c <= '9') { - factor *= 0.1; - value += (c - '0') * factor; - next_char(); - } + for (;;) { + switch (c) { + case SYMBOL_CHARS_E_P: + obstack_1grow(&symbol_obstack, (char)c); + next_char(); + if (c == '+' || c == '-') { + case '.': + case DIGITS: + case SYMBOL_CHARS_WITHOUT_E_P: + obstack_1grow(&symbol_obstack, (char)c); + next_char(); + } + break; - if(c == 'e' || c == 'E') { - next_char(); - parse_floatingpoint_exponent(value); - return; + default: +dollar_sign: + lexer_token.kind = T_NUMBER; + lexer_token.literal.string = sym_make_string(STRING_ENCODING_CHAR); + return; + } } - - lexer_token.type = T_FLOATINGPOINT; - lexer_token.v.floatvalue = value; - - parse_floating_suffix(); } -static void parse_number_dec(void) +/** + * Returns true if the given char is a octal digit. + * + * @param char the character to check + */ +static bool is_octal_digit(utf32 chr) { - int value = 0; - - while(isdigit(c)) { - value = 10 * value + c - '0'; - next_char(); - } - - if(c == '.') { - next_char(); - parse_floatingpoint_fract(value); - return; - } - if(c == 'e' || c == 'E') { - next_char(); - parse_floatingpoint_exponent(value); - return; - } - parse_integer_suffix(); - - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = value; + return '0' <= chr && chr <= '7'; } -static void parse_number(void) +/** + * Returns the value of a digit. + * The only portable way to do it ... + */ +static int digit_value(utf32 const digit) { - if (c == '0') { - next_char(); - switch (c) { - case 'X': - case 'x': - parse_number_hex(); - break; - case '0': - case '1': - case '2': - case '3': - case '4': - case '5': - case '6': - case '7': - parse_number_oct(); - break; - case '.': - next_char(); - parse_floatingpoint_fract(0); - break; - case 'e': - case 'E': - parse_floatingpoint_exponent(0); - break; - case '8': - case '9': - next_char(); - parse_error("invalid octal number"); - lexer_token.type = T_ERROR; - return; - default: - put_back(c); - c = '0'; - parse_number_dec(); - return; - } - } else { - parse_number_dec(); + switch (digit) { + case '0': return 0; + case '1': return 1; + case '2': return 2; + case '3': return 3; + case '4': return 4; + case '5': return 5; + case '6': return 6; + case '7': return 7; + case '8': return 8; + case '9': return 9; + case 'a': + case 'A': return 10; + case 'b': + case 'B': return 11; + case 'c': + case 'C': return 12; + case 'd': + case 'D': return 13; + case 'e': + case 'E': return 14; + case 'f': + case 'F': return 15; + default: + internal_error("wrong character given"); } } -static int parse_octal_sequence(void) +/** + * Parses an octal character sequence. + * + * @param first_digit the already read first digit + */ +static utf32 parse_octal_sequence(utf32 const first_digit) { - int value = 0; - while(1) { - if(c < '0' || c > '7') - break; - value = 8 * value + c - '0'; - next_char(); - } - + assert(is_octal_digit(first_digit)); + utf32 value = digit_value(first_digit); + if (!is_octal_digit(c)) return value; + value = 8 * value + digit_value(c); + next_char(); + if (!is_octal_digit(c)) return value; + value = 8 * value + digit_value(c); + next_char(); return value; } -static int parse_hex_sequence(void) +/** + * Parses a hex character sequence. + */ +static utf32 parse_hex_sequence(void) { - int value = 0; - while(1) { - if (c >= '0' && c <= '9') { - value = 16 * value + c - '0'; - } else if ('A' <= c && c <= 'F') { - value = 16 * value + c - 'A' + 10; - } else if ('a' <= c && c <= 'f') { - value = 16 * value + c - 'a' + 10; - } else { - break; - } + utf32 value = 0; + while (isxdigit(c)) { + value = 16 * value + digit_value(c); next_char(); } - return value; } -static int parse_escape_sequence(void) +/** + * Parse an escape sequence. + */ +static utf32 parse_escape_sequence(void) { eat('\\'); - int ec = c; + utf32 const ec = c; next_char(); - switch(ec) { + switch (ec) { case '"': return '"'; - case '\'': return'\''; + case '\'': return '\''; case '\\': return '\\'; case '?': return '\?'; case 'a': return '\a'; @@ -536,150 +558,128 @@ static int parse_escape_sequence(void) case '5': case '6': case '7': - return parse_octal_sequence(); + return parse_octal_sequence(ec); case EOF: parse_error("reached end of file while parsing escape sequence"); return EOF; + /* \E is not documented, but handled, by GCC. It is acceptable according + * to §6.11.4, whereas \e is not. */ + case 'E': + case 'e': + if (c_mode & _GNUC) + return 27; /* hopefully 27 is ALWAYS the code for ESCAPE */ + break; + + case 'U': return parse_universal_char(8); + case 'u': return parse_universal_char(4); + default: - parse_error("unknown escape sequence"); - return EOF; + break; } + /* §6.4.4.4:8 footnote 64 */ + parse_error("unknown escape sequence"); + return EOF; } -const char *concat_strings(const char *s1, const char *s2) +string_t make_string(const char *string) { - size_t len1 = strlen(s1); - size_t len2 = strlen(s2); - - char *concat = obstack_alloc(&symbol_obstack, len1 + len2 + 1); - memcpy(concat, s1, len1); - memcpy(concat + len1, s2, len2 + 1); - - const char *result = strset_insert(&stringset, concat); - if(result != concat) { - obstack_free(&symbol_obstack, concat); - } - - return result; + obstack_grow(&symbol_obstack, string, strlen(string)); + return sym_make_string(STRING_ENCODING_CHAR); } -static void parse_string_literal(void) +static void parse_string(utf32 const delim, token_kind_t const kind, string_encoding_t const enc, char const *const context) { - unsigned start_linenr = lexer_token.source_position.linenr; - char *string; - const char *result; + eat(delim); - assert(c == '"'); - next_char(); + while (true) { + switch (c) { + case '\\': { + utf32 const tc = parse_escape_sequence(); + if (enc == STRING_ENCODING_CHAR) { + if (tc >= 0x100) { + warningf(WARN_OTHER, &lexer_pos, "escape sequence out of range"); + } + obstack_1grow(&symbol_obstack, tc); + } else { + obstack_grow_utf8(&symbol_obstack, tc); + } + break; + } - int tc; - while(1) { - switch(c) { - case '\\': - tc = parse_escape_sequence(); - obstack_1grow(&symbol_obstack, tc); + case NEWLINE: + errorf(&lexer_pos, "newline while parsing %s", context); break; case EOF: - error_prefix_at(lexer_token.source_position.input_name, - start_linenr); - fprintf(stderr, "string has no end\n"); - lexer_token.type = T_ERROR; - return; - - case '"': - next_char(); + errorf(&lexer_token.base.source_position, "EOF while parsing %s", context); goto end_of_string; default: - obstack_1grow(&symbol_obstack, c); - next_char(); - break; + if (c == delim) { + next_char(); + goto end_of_string; + } else { + obstack_grow_utf8(&symbol_obstack, c); + next_char(); + break; + } } } end_of_string: - - /* TODO: concatenate multiple strings separated by whitespace... */ - - /* add finishing 0 to the string */ - obstack_1grow(&symbol_obstack, '\0'); - string = obstack_finish(&symbol_obstack); - - /* check if there is already a copy of the string */ - result = strset_insert(&stringset, string); - if(result != string) { - obstack_free(&symbol_obstack, string); - } - - lexer_token.type = T_STRING_LITERAL; - lexer_token.v.string = result; + lexer_token.kind = kind; + lexer_token.literal.string = sym_make_string(enc); } -static void parse_character_constant(void) +/** + * Parse a string literal and set lexer_token. + */ +static void parse_string_literal(string_encoding_t const enc) { - eat('\''); - - int found_char = 0; - while(1) { - switch(c) { - case '\\': - found_char = parse_escape_sequence(); - break; - - MATCH_NEWLINE( - parse_error("newline while parsing character constant"); - break; - ) - - case '\'': - next_char(); - goto end_of_char_constant; - - case EOF: - parse_error("EOF while parsing character constant"); - lexer_token.type = T_ERROR; - return; + parse_string('"', T_STRING_LITERAL, enc, "string literal"); +} - default: - if(found_char != 0) { - parse_error("more than 1 characters in character " - "constant"); - goto end_of_char_constant; - } else { - found_char = c; - next_char(); - } - break; - } +/** + * Parse a character constant and set lexer_token. + */ +static void parse_character_constant(string_encoding_t const enc) +{ + parse_string('\'', T_CHARACTER_CONSTANT, enc, "character constant"); + if (lexer_token.literal.string.size == 0) { + errorf(&lexer_token.base.source_position, "empty character constant"); } - -end_of_char_constant: - lexer_token.type = T_INTEGER; - lexer_token.v.intvalue = found_char; } +/** + * Skip a multiline comment. + */ static void skip_multiline_comment(void) { - unsigned start_linenr = lexer_token.source_position.linenr; - - while(1) { - switch(c) { + while (true) { + switch (c) { + case '/': + next_char(); + if (c == '*') { + /* nested comment, warn here */ + warningf(WARN_COMMENT, &lexer_pos, "'/*' within comment"); + } + break; case '*': next_char(); - if(c == '/') { + if (c == '/') { next_char(); return; } break; - MATCH_NEWLINE(break;) + case NEWLINE: + break; - case EOF: - error_prefix_at(lexer_token.source_position.input_name, - start_linenr); - fprintf(stderr, "at end of file while looking for comment end\n"); + case EOF: { + errorf(&lexer_token.base.source_position, + "at end of file while looking for comment end"); return; + } default: next_char(); @@ -688,10 +688,13 @@ static void skip_multiline_comment(void) } } +/** + * Skip a single line comment. + */ static void skip_line_comment(void) { - while(1) { - switch(c) { + while (true) { + switch (c) { case EOF: return; @@ -699,6 +702,14 @@ static void skip_line_comment(void) case '\r': return; + case '\\': + next_char(); + if (c == '\n' || c == '\r') { + warningf(WARN_COMMENT, &lexer_pos, "multi-line comment"); + return; + } + break; + default: next_char(); break; @@ -706,115 +717,175 @@ static void skip_line_comment(void) } } +/** The current preprocessor token. */ static token_t pp_token; +/** + * Read the next preprocessor token. + */ static inline void next_pp_token(void) { lexer_next_preprocessing_token(); pp_token = lexer_token; } +/** + * Eat all preprocessor tokens until newline. + */ static void eat_until_newline(void) { - while(pp_token.type != '\n' && pp_token.type != T_EOF) { + while (pp_token.kind != '\n' && pp_token.kind != T_EOF) { next_pp_token(); } } -static void error_directive(void) -{ - error_prefix(); - fprintf(stderr, "#error directive: \n"); - - /* parse pp-tokens until new-line */ -} - -static void define_directive(void) +/** + * Parse the line directive. + */ +static void parse_line_directive(void) { - lexer_next_preprocessing_token(); - if(lexer_token.type != T_IDENTIFIER) { - parse_error("expected identifier after #define\n"); - eat_until_newline(); + if (pp_token.kind != T_NUMBER) { + parse_error("expected integer"); + } else { + /* use offset -1 as this is about the next line */ + char *end; + long const line = strtol(pp_token.literal.string.begin, &end, 0); + if (*end == '\0') { + lexer_pos.lineno = line - 1; + } else { + errorf(&lexer_pos, "'%S' is not a valid line number", &pp_token.literal.string); + } + next_pp_token(); } -} + if (pp_token.kind == T_STRING_LITERAL && pp_token.literal.string.encoding == STRING_ENCODING_CHAR) { + lexer_pos.input_name = pp_token.literal.string.begin; + lexer_pos.is_system_header = false; + next_pp_token(); -static void ifdef_directive(int is_ifndef) -{ - (void) is_ifndef; - lexer_next_preprocessing_token(); - //expect_identifier(); - //extect_newline(); -} + /* attempt to parse numeric flags as outputted by gcc preprocessor */ + while (pp_token.kind == T_NUMBER) { + /* flags: + * 1 - indicates start of a new file + * 2 - indicates return from a file + * 3 - indicates system header + * 4 - indicates implicit extern "C" in C++ mode + * + * currently we're only interested in "3" + */ + if (streq(pp_token.literal.string.begin, "3")) { + lexer_pos.is_system_header = true; + } + next_pp_token(); + } + } -static void endif_directive(void) -{ - //expect_newline(); + eat_until_newline(); } -static void parse_line_directive(void) +/** + * STDC pragmas. + */ +typedef enum stdc_pragma_kind_t { + STDC_UNKNOWN, + STDC_FP_CONTRACT, + STDC_FENV_ACCESS, + STDC_CX_LIMITED_RANGE +} stdc_pragma_kind_t; + +/** + * STDC pragma values. + */ +typedef enum stdc_pragma_value_kind_t { + STDC_VALUE_UNKNOWN, + STDC_VALUE_ON, + STDC_VALUE_OFF, + STDC_VALUE_DEFAULT +} stdc_pragma_value_kind_t; + +/** + * Parse a pragma directive. + */ +static void parse_pragma(void) { - if(pp_token.type != T_INTEGER) { - parse_error("expected integer"); - } else { - lexer_token.source_position.linenr = pp_token.v.intvalue - 1; - next_pp_token(); + next_pp_token(); + if (pp_token.kind != T_IDENTIFIER) { + warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position, + "expected identifier after #pragma"); + eat_until_newline(); + return; } - if(pp_token.type == T_STRING_LITERAL) { - lexer_token.source_position.input_name = pp_token.v.string; + + stdc_pragma_kind_t kind = STDC_UNKNOWN; + if (pp_token.base.symbol->pp_ID == TP_STDC && c_mode & _C99) { + /* a STDC pragma */ next_pp_token(); - } + switch (pp_token.base.symbol->pp_ID) { + case TP_FP_CONTRACT: kind = STDC_FP_CONTRACT; break; + case TP_FENV_ACCESS: kind = STDC_FENV_ACCESS; break; + case TP_CX_LIMITED_RANGE: kind = STDC_CX_LIMITED_RANGE; break; + default: break; + } + if (kind != STDC_UNKNOWN) { + next_pp_token(); + stdc_pragma_value_kind_t value; + switch (pp_token.base.symbol->pp_ID) { + case TP_ON: value = STDC_VALUE_ON; break; + case TP_OFF: value = STDC_VALUE_OFF; break; + case TP_DEFAULT: value = STDC_VALUE_DEFAULT; break; + default: value = STDC_VALUE_UNKNOWN; break; + } + if (value == STDC_VALUE_UNKNOWN) { + kind = STDC_UNKNOWN; + errorf(&pp_token.base.source_position, "bad STDC pragma argument"); + } + } + } eat_until_newline(); + if (kind == STDC_UNKNOWN) { + warningf(WARN_UNKNOWN_PRAGMAS, &pp_token.base.source_position, + "encountered unknown #pragma"); + } } +/** + * Parse a preprocessor non-null directive. + */ static void parse_preprocessor_identifier(void) { - assert(pp_token.type == T_IDENTIFIER); - symbol_t *symbol = pp_token.v.symbol; - - switch(symbol->pp_ID) { - case TP_include: - printf("include - enable header name parsing!\n"); - break; - case TP_define: - define_directive(); - break; - case TP_ifdef: - ifdef_directive(0); - break; - case TP_ifndef: - ifdef_directive(1); - break; - case TP_endif: - endif_directive(); - break; + assert(pp_token.kind == T_IDENTIFIER); + switch (pp_token.base.symbol->pp_ID) { case TP_line: next_pp_token(); parse_line_directive(); break; - case TP_if: - case TP_else: - case TP_elif: - case TP_undef: - case TP_error: - error_directive(); - break; case TP_pragma: + parse_pragma(); + break; + case TP_error: + /* TODO; output the rest of the line */ + parse_error("#error directive"); break; } } -static void parse_preprocessor_directive() +/** + * Parse a preprocessor directive. + */ +static void parse_preprocessor_directive(void) { next_pp_token(); - switch(pp_token.type) { + switch (pp_token.kind) { case T_IDENTIFIER: parse_preprocessor_identifier(); break; - case T_INTEGER: + case T_NUMBER: parse_line_directive(); break; + case '\n': + /* NULL directive, see §6.10.7 */ + break; default: parse_error("invalid preprocessor directive"); eat_until_newline(); @@ -824,68 +895,93 @@ static void parse_preprocessor_directive() #define MAYBE_PROLOG \ next_char(); \ - while(1) { \ - switch(c) { + while (true) { \ + switch (c) { #define MAYBE(ch, set_type) \ case ch: \ next_char(); \ - lexer_token.type = set_type; \ + lexer_token.kind = set_type; \ return; +/* must use this as last thing */ +#define MAYBE_MODE(ch, set_type, mode) \ + case ch: \ + if (c_mode & mode) { \ + next_char(); \ + lexer_token.kind = set_type; \ + return; \ + } \ + /* fallthrough */ + #define ELSE_CODE(code) \ default: \ - code; \ + code \ + return; \ } \ - } /* end of while(1) */ \ - break; + } /* end of while (true) */ \ #define ELSE(set_type) \ ELSE_CODE( \ - lexer_token.type = set_type; \ - return; \ + lexer_token.kind = set_type; \ ) void lexer_next_preprocessing_token(void) { - while(1) { - switch(c) { + while (true) { + lexer_token.base.source_position = lexer_pos; + lexer_token.base.symbol = NULL; + + switch (c) { case ' ': case '\t': next_char(); break; - MATCH_NEWLINE( - lexer_token.type = '\n'; + case NEWLINE: + lexer_token.kind = '\n'; return; - ) - SYMBOL_CHARS + case SYMBOL_CHARS: { parse_symbol(); + /* might be a wide string ( L"string" ) */ + string_encoding_t const enc = STRING_ENCODING_WIDE; + if (lexer_token.base.symbol == symbol_L) { + switch (c) { + case '"': parse_string_literal(enc); break; + case '\'': parse_character_constant(enc); break; + } + } return; + } - DIGITS - parse_number(); + case DIGITS: + parse_pp_number(); return; case '"': - parse_string_literal(); + parse_string_literal(STRING_ENCODING_CHAR); return; case '\'': - parse_character_constant(); + parse_character_constant(STRING_ENCODING_CHAR); return; case '.': MAYBE_PROLOG + case DIGITS: + put_back(c); + c = '.'; + parse_pp_number(); + return; + case '.': MAYBE_PROLOG MAYBE('.', T_DOTDOTDOT) ELSE_CODE( put_back(c); c = '.'; - lexer_token.type = '.'; - return; + lexer_token.kind = '.'; ) ELSE('.') case '&': @@ -928,25 +1024,24 @@ void lexer_next_preprocessing_token(void) ELSE('/') case '%': MAYBE_PROLOG - MAYBE('>', T_PERCENTGREATER) + MAYBE('>', '}') MAYBE('=', T_PERCENTEQUAL) case ':': MAYBE_PROLOG case '%': MAYBE_PROLOG - MAYBE(':', T_PERCENTCOLONPERCENTCOLON) + MAYBE(':', T_HASHHASH) ELSE_CODE( put_back(c); c = '%'; - lexer_token.type = T_PERCENTCOLON; - return; + lexer_token.kind = '#'; ) - ELSE(T_PERCENTCOLON) + ELSE('#') ELSE('%') case '<': MAYBE_PROLOG - MAYBE(':', T_LESSCOLON) - MAYBE('%', T_LESSPERCENT) + MAYBE(':', '[') + MAYBE('%', '{') MAYBE('=', T_LESSEQUAL) case '<': MAYBE_PROLOG @@ -972,7 +1067,8 @@ void lexer_next_preprocessing_token(void) ELSE('|') case ':': MAYBE_PROLOG - MAYBE('>', T_COLONGREATER) + MAYBE('>', ']') + MAYBE_MODE(':', T_COLONCOLON, _CXX) ELSE(':') case '=': MAYBE_PROLOG @@ -983,6 +1079,17 @@ void lexer_next_preprocessing_token(void) MAYBE('#', T_HASHHASH) ELSE('#') + case '\\': + next_char(); + if (c == 'U' || c == 'u') { + put_back(c); + c = '\\'; + parse_symbol(); + } else { + lexer_token.kind = '\\'; + } + return; + case '?': case '[': case ']': @@ -993,21 +1100,19 @@ void lexer_next_preprocessing_token(void) case '~': case ';': case ',': - case '\\': - lexer_token.type = c; + lexer_token.kind = c; next_char(); return; case EOF: - lexer_token.type = T_EOF; + lexer_token.kind = T_EOF; return; default: +dollar_sign: + errorf(&lexer_pos, "unknown character '%lc' found", c); next_char(); - error_prefix(); - fprintf(stderr, "unknown character '%c' found\n", c); - lexer_token.type = T_ERROR; - return; + break; } } } @@ -1015,15 +1120,13 @@ void lexer_next_preprocessing_token(void) void lexer_next_token(void) { lexer_next_preprocessing_token(); - if(lexer_token.type != '\n') - return; + while (lexer_token.kind == '\n') { newline_found: - do { lexer_next_preprocessing_token(); - } while(lexer_token.type == '\n'); + } - if(lexer_token.type == '#') { + if (lexer_token.kind == '#') { parse_preprocessor_directive(); goto newline_found; } @@ -1032,13 +1135,27 @@ newline_found: void init_lexer(void) { strset_init(&stringset); + symbol_L = symbol_table_insert("L"); } -void lexer_open_stream(FILE *stream, const char *input_name) +static void input_error(unsigned delta_lines, unsigned delta_cols, + const char *message) { - input = stream; - lexer_token.source_position.linenr = 0; - lexer_token.source_position.input_name = input_name; + lexer_pos.lineno += delta_lines; + lexer_pos.colno += delta_cols; + errorf(&lexer_pos, "%s", message); +} + +void lexer_switch_input(input_t *new_input, const char *input_name) +{ + lexer_pos.lineno = 0; + lexer_pos.colno = 0; + lexer_pos.input_name = input_name; + + set_input_error_callback(input_error); + input = new_input; + bufpos = NULL; + bufend = NULL; /* place a virtual \n at the beginning so the lexer knows that we're * at the beginning of a line */ @@ -1053,7 +1170,7 @@ void exit_lexer(void) static __attribute__((unused)) void dbg_pos(const source_position_t source_position) { - fprintf(stdout, "%s:%d\n", source_position.input_name, - source_position.linenr); + fprintf(stdout, "%s:%u:%u\n", source_position.input_name, + source_position.lineno, (unsigned)source_position.colno); fflush(stdout); }